A test to detect replication in HIV serological data labelled by birth date based on the number of matching pairs in a sample.

Diagnoses of HIV infection are reported to the Public Health Laboratory Service (PHLS) by microbiologists through a voluntary confidential surveillance scheme. Names are not recorded on the database but the date of birth of the individual concerned is usually available. This paper discusses a statistical method to detect repeated counting of individuals in these and similar data based on the number of matching pairs in the sample. The test is based on the theoretical result that the null hypothesis of all birth dates equally likely and all individuals distinct minimizes the expected number of matching pairs in the sample. Five of the 16 birth years in the sample taken in 1991 show evidence of more replication than would be expected by chance using a 5 per cent level test. When the test is repeated taking into account a small but statistically significant seasonal variation in the birth rate, the results are very similar.